Abstract: A method of estimating a number of agents needed in a queueing system where each active agent is available to work on several tasks in parallel, includes the steps of obtaining a representation of the queueing system comprising information on all possible states that the queueing system can be in and the intensity rates for the possible transitions between states calculating the resulting service level for the tentative number of active agents based on the state probability vector, which is based on the representation, a service rate values ?i,j and an arrival rate intensity value ? and comparing the resulting service level to the target service level.